Home Page

Online Casebook

Officers

How to Join

Return to RiskWorld

SRA Risk Science & Law Specialty Group

Merrell Dow Pharmaceuticals, Inc. v. Havner: Favoring Epidemiological Evidence

Does recovery of damages in a toxic tort lawsuit require evidence of statistically significant human (epidemiological) studies that demonstrate a positive association between a toxic agent and a toxic injury?

In 1994 the Texas Court of Appeals decided Merrell Dow Pharmaceuticals, Inc. v. Havner, 907 S.W.2d 535 (Tex.App.—Corpus Christi 1994). The appellate court had previously reversed a jury verdict because plaintiffs' experts could not cite one epidemiological study that found a statistically significant association between a suspect teratogen and the plaintiff's birth defects. On original submission a panel of this court found no evidence to support causation, but in a rehearing en banc (full court), the jury verdict was affirmed with respect to both admissibility and sufficiency of evidence for causation (Chief Justice dissenting). A changed interpretation of the epidemiological evidence followed these findings:

§ Biological causation is inherently difficult to assess based on epidemiological studies. That thalidomide causes limb reduction defects based solely on epidemiological studies cannot be proved although the scientific community accepts that it does. Only one drug (Accutane) has been proven teratogenic over the past 25 years through epidemiological studies alone.

§ None of the numerous negative epidemiological studies has the requisite size and statistical power to detect a change in the rate of limb reduction defects occurring by themselves and not as part of a cluster of birth abnormalities, nor do most of them specify when during the pregnancy the drug exposure occurred.

§ A reanalysis of the relative risk (RR) or association between the suspect drug and the plaintiff's birth defects that accounts for the mother's exposure during the first trimester rather than during the pregnancy as a whole, or at the time in gestation when fetal limb formation is critical, results in RR = 2.13, a positive association, instead of 0.97, which is inconclusive.

§ Reanalysis of epidemiological data and the resulting opinion of a scientist is not, as a matter of principle, insufficient to support a finding of causation in any case. Such reanalysis is a methodology that the FDA and EPA use to verify reports they get from the industries they regulate.

§ Epidemiology is considered a "soft" science because the researcher cannot experimentally control the variables, as compared to the "hard" science of laboratory animal testing. The EPA relies solely on animal studies to determine the effects of most chemicals, while the FDA relies heavily on animal studies in the early stages of drug approval.

The potential influence of this case on risk analysis may be to place the limitations of epidemiology in perspective so that many flawed human studies are no longer consistently preferred as evidence of causation over carefully designed and controlled animal studies.

[Contributed by Wayne Roth-Nelson]

CLASSIFICATION: CIVIL LITIGATION; TOXIC CHEMICAL INJURY; SUFFICIENCY OF EVIDENCE FOR CAUSATION

Copyright © 1999 by Tec-Com, Inc.
Last modified July 29, 1999.